first commit

app.py

@@ -1,13 +1,12 @@
-import spaces
 import gradio as gr
 from transformers import pipeline
 import torch
 import os
+import spaces
 
 hf_token = os.environ["HF_TOKEN"]
 
 # Load the Gemma 3 pipeline.
-# Gemma 3 is a multimodal model that accepts text and image inputs.
 pipe = pipeline(
     "image-text-to-text",
     model="google/gemma-3-4b-it",
@@ -16,53 +15,44 @@ pipe = pipeline(
     use_auth_token=hf_token
 )
 @spaces.GPU
-def generate_response(user_text, user_image, history):
+def generate_response(user_text, user_image):
+    # Check if an image was uploaded.
+    if user_image is None:
+        return "Error: An image upload is mandatory."
+    
+    # Prepare messages with the system prompt and user inputs.
     messages = [
         {
             "role": "system",
             "content": [{"type": "text", "text": "You are a helpful assistant."}]
         }
     ]
-    user_content = []
-    if user_image is not None:
-        user_content.append({"type": "image", "image": user_image})
+    user_content = [{"type": "image", "image": user_image}]
     if user_text:
         user_content.append({"type": "text", "text": user_text})
     messages.append({"role": "user", "content": user_content})
     
-    # Call the pipeline with the provided messages.
+    # Call the pipeline.
     output = pipe(text=messages, max_new_tokens=200)
-
-    print(output)
-    print(output[0]["generated_text"][-1]["content"])
     
-    # Attempt to extract the generated content using the expected structure.
+    # Try to extract the generated content.
     try:
         response = output[0]["generated_text"][-1]["content"]
-        history.append((user_text, response))
-
     except (KeyError, IndexError, TypeError):
-        # Fallback: return the raw output as a string.
-        #print(response)
-        pass
-        #response = str(output)
+        response = str(output)
     
-    return history, history 
-
-with gr.Blocks() as demo:
-    gr.Markdown("# Gemma 3 Chat Interface")
-    gr.Markdown(
-        "This interface lets you chat with the Gemma 3 model. "
-        "You can type a message and optionally attach an image."
-    )
-    # Specify type="messages" to avoid deprecation warnings.
-    chatbot = gr.Chatbot(type="messages")
-    with gr.Row():
-        txt = gr.Textbox(show_label=False, placeholder="Type your message here...", container=False)
-        img = gr.Image(type="pil", label="Attach an image (optional)")
-    state = gr.State([])
+    return response
 
-    txt.submit(generate_response, inputs=[txt, img, state], outputs=[chatbot, state])
+iface = gr.Interface(
+    fn=generate_response,
+    inputs=[
+        gr.Textbox(label="Message", placeholder="Type your message here..."),
+        gr.Image(type="pil", label="Upload an Image", source="upload")
+    ],
+    outputs=gr.Textbox(label="Response"),
+    title="Gemma 3 Simple Interface",
+    description="Enter your message and upload an image (image upload is mandatory) to get a response."
+)
 
 if __name__ == "__main__":
-    demo.launch()
+    iface.launch()